Problems in shifting the ground of a power supply

Hi all,

In order to measure the low noise of a device at a negative voltage between -5V and -1V, I developed the following circuit:

- Transimpedance amplifier using OPA656U from Texas instruments that functions at Vs=-/+5V.
- A circuit with two voltage regulators (negative and positive) that regulate from -/+12V to -/+5V.
- Batteries
- Board for the ground separation

The purpose is to shift the ground of the power supply to the voltage of the Keithley230’s (K230’s) and let –Vs and +Vs follow that shift by adding their voltage values to -/+5V.

In order to do this, we separated all the grounds of the board with the amplifier from the ground of the power supply that will be on another potential (the potential of K230). Moreover, we disconnected the ground of the batteries from the regulators and connected it to the ground of the K230.

Thus, the results obtained for a voltage of -3V for example on the K230 must be on the power supply as follows :
-Vs = -5-3 = -8V
Gnd = -3V
+Vs = +5-3 = +2V

However, the results obtained are :
-Vs = -5V
Gnd = -3V
+Vs = +5V

The circuit looks correct on paper but when developing it, it is not working.
Any idea about the strange values obtained ?

Kind regards,
SKEB

It's too hard for my poor brain to follow your description of your circuit. Please post a circuit diagram.

I tried to attach the circuit in the first message but it seems it didnt work.

Please find it attached in the following message. (I hope it works)

Where is the ground of the K230 connected?

I still confused as to what you are trying to do. :?: Why do you want to shift the power supply voltages?

If you are measuring AC noise, can't you just couple the signal through a capacitor to block the DC?

Breaking the connection between the batteries common and the voltage regulators ground will not work for several possible reasons. You may not get +/-5V or perhaps one of the regulators may not start. Since you are using batteries, breaking the connection may not be necessary. However, more detail is needed. Can you provide more detail of the Keithley and signal source connections? Is the Keithley a sensor bias, etc.? Also include how you are measuring the amplifier output. What is the instrumentation? How is it connected. Enough detail is needed to show why you are trying to float the circuit. Your diagram is a good start.

Hi and thank you for your answers.

I solved my problem and I want to answer your questions.

Why I am shifting the ground into the K230's voltage ?
Because the output voltage swing of the OPA656 is around -/+ 3V. This way I put the ground on the K230 voltage (that is on the non-inverting input of the OPA656) and the power supply varies relatively to the new ground voltage. Thus I have a larger range of input positive and negative voltages.

How I am measuring the amplifier output. What is the instrumentation?
On the output I have a NI PCI 4461 card for National instruments connected with a BNC connector.

I am in AC coupled configuration, but I have a transimpedance amplifier (OPA656 with a feedback resistor). I think that I don't need the coupling capacitor since I am feeding a current.

Thank you !